The Journal

of Neuroscience,

January

1990,

10(l):

247-255

Local Accumulation of Acetylcholine Receptors Is Neither Necessary Nor Sufficient to Induce Cluster Formation Jes Stollberg Department

and Scott E. Fraser of Physiology and Biophysics,

College of Medicine, University

Acetylcholine receptors (AChRs) accumulate at developing neuromuscular junctions in part via lateral migration of diffusely expressed receptors. Using a model system-cultured Xenopus muscle cells exposed to electric fields-we have shown that AChRs, concentrated at the cathode-facing cell pole, continue to aggregate there after the field is terminated (Stollberg and Fraser, 1988). These observations are consistent with the possibility that the field-induced increase in receptor concentration triggers the aggregation event. Only 2 other molecular events could initiate the electric field-induced receptor aggregation: (1) a local increase in the density of some other molecules, or (2) a voltagesensitive mechanism. Treatment of muscle cell cultures with neuraminidase changes the cell surface charge and has been reported to reverse the direction of electromigration for AChRs and concanavalin A binding sites (Orida and Poo, 1978). Using digitally analyzed fluorescence videomicroscopy, we find that AChRs in neuraminidase-treated cultures accumulate at both cell poles in an electric field. After termination of the field, the AChR continues to aggregate at the cathode-facing pole, as in cells not treated with neuraminidase. However, receptor density decreases at the anode-facing pole, indicating that elevated AChR density does not initiate receptor aggregation. Cells pretreated with neuraminidase and trypsin (which blocks receptor aggregation) display reversed receptor distributions compared to untreated controls, indicating that electromigration has indeed been reversed. The rate at which neuraminidaseand trypsin-treated cells approach steadystate distributions indicates a receptor diffusion constant of = 1.2 x 1 O-s cm%ec, consistent with a diffusion trap mechanism of receptor aggregation. These results are the first conclusive demonstration that the local concentration of receptors is neither necessary nor sufficient to induce receptor clustering. Our observations suggest that receptor clustering is triggered by the accumulation of some other molecules, or by a voltage-sensitive mechanism.

An important problem in developmental neurobiology concerns the localization of specific moleculesat synapses.A much studied example is the neuromuscularjunction, at which acetylchoReceived May 3, 1989; revised July 21, 1989; accepted July 25, 1989. This work was supported by a grant from the Monsanto Corporation and by N.I.H. Training Grant HD07029. Correspondence should be addressed to Jes Stollberg at the above address. Copyright 0 1990 Society for Neuroscience 0270-6474/90/010247-09$02.00/O

of California, Irvine, Irvine, California 92717

line receptors(AChRs), acetylcholinesterase,basallamina components, and numerous cytoskeletal elementsare concentrated (for review seeDennis, 1981; Schuetze and Role, 1987). Prior to innervation, the AChRs are distributed diffusely; clustering begins shortly after neuronal contact both in vivo (Blackshaw and Warner, 1976; Creazzoand Sohal, 1983)and in vitro (Frank and Fischbach, 1979; Kidokoro et al., 1980; Role et al., 1987). The development and stabilization of receptor clusterscontinues over a period of days to weeks and is likely to involve multiple mechanismsacting in concert as the cluster matures. It is of crucial importance, therefore, to considerthe time frame of experiments directed at questionsof mechanism.The focus of this report is on the initial events (i.e., within the first few minutes) responsiblefor triggering receptor clustering. Experiments following the distribution of labeledAChRs indicate that lateral migration of diffusely expressedreceptors makes a significant contribution to the initial clustering event (Anderson and Cohen, 1977; Frank and Fischbach, 1979; Kuromi and Kidokoro, 1984; Ziskind-Conhaim et al., 1984; Role et al., 1985; Stollberg and Fraser, 1988). In Xenopus, this migration has been shown to be consistent with the action of a diffusion trap mechanism(Edwardsand Frisch, 1976),in which passively diffusing receptors are locally immobilized (Kuromi and Kidokoro, 1984; Stollberg and Fraser, 1988).The selective trapping of AChRs and other synaptic componentsis presumably mediated by binding to membrane components,to extracellular matrix molecules,or to cytoskeletalelements.The initial events that trigger the relevant interactions remain to be elucidated. The clustering of membrane componentscan be studied by monitoring their redistribution on cultured cells in responseto externally applied electric fields (Jaffe, 1977; Poo and Robinson, 1977; Poo et al., 1978; Luther and Peng, 1985). The technique permits an experimental manipulation of molecular distributions that is independent of exogenousfactors and cell contacts, thereby facilitating quantitative analysesof receptor clustering. This approach hasbeenusedto particular advantagewith spherical musclecell (mysophere)cultures, as the geometry of these cells simplifiesthe testing of theoretical predictions concerning the electromigration, diffusion, and aggregationof concanavalin A (con A) binding sitesand AChRs (Orida and Poo, 1978, 1980, 1981; Poo et al., 1979; McLaughlin and Poo, 1981; Stollberg and Fraser, 1988). We have shownpreviously that cultured Xenopusmyospheres preserve at least someof the components required for the induction of receptor clustering (Stollberg and Fraser, 1988). AChRs accumulate at the cathode-facing cell pole in response to electric fields and continue to aggregateat that pole after the

248

.g

Stollberg

and

Fraser

- AChR

Accumulation

Does

Not Trigger

Clustering

01 0

30

60 Angle

90 from

Anode

120

150

180

(degrees)

Figure 1. Distributionof AChRs and con A sites following an electric field of 8 V/cm for 40 min. Data are presented as means + standard errors of site density vs the angle (0) from the anodal cell pole (see inset).

In thesecontrol experiments,cellswerenot enzymaticallypretreated. Circles, AChR distribution, n = I cells. Arrow, Location of minimum AChR densitv. Curve. Theoretical fit to ideal distribution with m/D = 577 + 36 V-l: _f, = 0.10 f 0.005; x2,,,, ~..,= 141, indicating that this curve does not adequately represent the data. Triangles, Con A site distribution. n = 11 cells. Curve. Theoretical fit with m/D = 28 2 2 V-l. f, = 0.68 + 0.05; xzcl,, = 11.’ This curve gives a good approximation “of the observed distribution. Inset, Analysis of cell images. The perimeter was divided into 16 sectors according to the angle with respect to the electric field (the sectors are mirror-symmetric about the field axis). The average receptor density for the distribution shown is loo%, although it may appear to be higher. This is because the sectors near the cell poles (0, 180”) represent lessof the spheresurfaceareathan thosenear the equator. In order to calculate the total number of receptors on the cell surface,the intensityfound at eachsectoris scaledby its surface area. Therefore, in calculating the average receptor density, the sectors

nearthe polesaregivenlessweight. field has been terminated.’

The aggregation

of receptors

is spe-

cific; con A binding sitesand lipids, which accumulate at the cathodal pole during exposure to a field, return to a uniform distribution after termination of the field (Poo et al., 1979; McLaughlin and Poo, 1981; Stollberg and Fraser, 1988). Receptor clustering in this system is consistent with a diffusion trap mechanism,is sensitive to trypsin digestion, and is insensitive to agents disrupting microtubules and microfilaments (Stollberg and Fraser, 1988). Together theseresultssuggestthat the aggregationof AChRs is mediated by adhesionor cohesion eventson the extracellular face of the membrane.The simplicity and rapid responseof the system(receptor clusteringis triggered within 5 min) render this a powerful experimental approach with which to probe the molecular events responsiblefor receptor clustering. A simple and attractive model accounting for these observations is that the field-induced accumulation of AChRs at the cathodal pole triggersreceptor aggregation.Under sucha model, increasedlocal density of receptorswould play a causalrole by shifting the equilibrium binding of receptorsto one another, or to components in or adjacent to the cell membrane. This hypothesisis consistent with observations of AChR cluster formation in aneuralcultures (Pumplin and Bloch, 1987),and with a widely held theory of patching and capping (Bourguignon and Bourguignon, 1984;seealsoDiscussion).The rigorous exclusion ’ We use “accumulation” or “concentration” to indicate a reversible increase in the density of a membrane component-due in our experiments to the application of electric fields. “Aggregation” or “clustering” is used to designate an irreversible association of membrane components.

of the hypothesisrequiresfinding experimental conditions under which the aggregationof receptors can be dissociatedfrom an antecedent increasein receptor density. The searchfor suchexperimental conditions requiresan understandingof the interactions betweenmembranecomponents and electric fields. The driving force for the field-induced redistribution of membrane components (“electromigration”) is thought to be a combination of electrophoresisand electroosmosis (McLaughlin and Poo, 1981). Electroosmosiscan be understood qualitatively by considering the negative charge bound to the extracellular membranesurface.As a consequence of this net negativity, positive ions in the saline are drawn to the membrane surface. These ions are highly mobile and in responseto an electric field will induce a “solvent drag” in the direction of the cathodal pole. This drag, or electroosmoticforce, acts on the extracellular aspectof membrane-boundmolecules, pushing them in the cathodal direction. In this way molecules that carry a large amount of negative charge (relative to the averagecell surfacecharge-the zeta potential) will migrate toward the anodal pole becausethe balanceof the 2 forces favors electrophoresis.Other moleculeswith a lesseramount of negative charge will

electromigrate

toward

the cathodal

cell pole

becauseelectroosmosisdominates.Evidence for the importance of the cell surface charge comes from experiments with neuraminidase, which removes negatively charged sialic acid residuesfrom the cell surface.This reducesthe magnitude of electroosmosis and should reverse the electromigration of some speciesfrom cathode-seekingto anode-seeking.The observed reversal of electromigration after such treatments supportsthe view that electroosmosisplays a significant role in electromigration (Orida and Poo, 1978; McLaughlin and Poo, 1981). Manipulation of electromigration by neuraminidase,in combination with the spatial resolution of the imaging techniques used here, offers a possible meansto separatereceptor accumulation and receptor aggregationexperimentally. The question addressedis whether the reversal of field-induced receptor migration is accompaniedby reversed receptor aggregation.Our results indicate that AChRs become clustered at the cathodal pole (asin cellsnot treated with neuraminidase)despitereversal of receptor electromigration. Therefore, elevated receptor density doesnot trigger the aggregationof receptors.

Materials

and Methods

Culture system. Myotomal cells were dissected from stage 18-20 Xenopus laevis embryos (Nieuwkoop and Faber, 1962) in Steinberg’s solution,

pH

7.8, containing

1 mg/ml

collagenase.

The cells were

disso-

ciated in Ca2+/MgZ+-free Steinberg’s solution, pH 7.8, and maintained on sterile coverslips in drops of culture medium (85% Steinberg’s solution, 10% Leibovitz’s L- 15 medium, 5% fetal calf serum, 50 &ml gentamicin), pH 7.8, for 1 d at 24°C. For details of culture system, field

application,anddataanalysisseeStollbergandFraser(1988).

Experimental manipulations. Prior to the application of electric fields, cultures were either not treated in any way (controls), incubated with neuraminidase, or incubated with neuraminidase followed by trypsin. Incubations were carried out at room temperature in 50 ~1 of Steinberg’s solution at pH 6.6 (neuraminidase) or pH 7.8 (trypsin). The cultures were

then

returned

to medium

and

assembled

into

electrophoresis

chambers fashioned from a microscope slide, coverslip runners, and the coverslip culture (internal chamber dimensions 6.0 cm x 1.0 cm x 0.2 mm deep). Electric potentials from an electrophoresis power supply were applied to the ends of the chambers via U-shaped 6-mm (I.D.) glass tubes filled with 2% agar/Steinberg’s solution. Separate agar bridges were used to monitor the potential drop across the chambers, which offers a direct measure of the field strength. Following field application, and the indicated post-field relaxation periods (if any) at room temperature, the cells were chilled and labeled for 10 min with 300 nM

The Journal of Neuroscience,

January

1990, 10(l)

249

Figure 2. Representative fluorescence videomicrographs of cells labeled with TMR-cr-Bgt. These images are examples of the raw data and have not been corrected for fluorescence background or camera nonlinearities. The videomicrographs illustrate the visually modest changes in receptor distribution seen under the various conditions, while the more sensitive quantitative analyses make use of both the above-mentioned corrections and the averaging of many cell images (Figs. 1, 3-7). A, Receptor distribution on a cell treated with 0.2 U/ml neuraminidase for 40 min and subjected to 8 V/cm for 80 min. Note the elevation of receptor density at both anode- and cathode-facing cell regions. B, As in A, but the cell was given 80 min of relaxation following field termination. Receptor aggregation has proceeded at the cathodal pole, while the density at the anodal pole has decreased. C, Receptor distribution on a cell pretreated with 0.2 U/ml neuraminidase for 40 min, then with 0.1% trypsin for 20 min, and then subjected to 8 V/cm for 40 min. Note the accumulation of receptors at the anode-facing pole and decrease at the cathodefacing pole. D, As in C, but the cell was given 80 min of relaxation following field termination. The receptor distribution is nearly uniform. The prominent dark spots in these images (,4-D) result from flaws in the camera tube; these along with other faulty pixels were rejected in the analysis. The bright regions in the cell interiors are due to autofluorescence (they appear even in unlabeled cells) and do not affect the quantitative image analysis. The field cathode is to the right in all cases. Calibration bar, 20 pm.

rhodamine-labeled cy-Bgt (TMR-or-Bgt), 25 &ml fluorescein-labeled concanavalin A (FLR-con A) in medium. The cultures were then rinsed and kept on ice until video images were acquired. Data analysis. Cell images were gathered through a Zeiss Universal microscope, collected by a SIT video camera (RCA model No. TC 1030). and stored on video cassettes (Sony U-ma&c video cassette recorder models VO-5600 and VO-5800). Videotane images were diaitized usine a D§or DS-88 board (Microworks). -Digital-images we;e corrected on a pixel-by-pixel basis for spatial aberrations in the illumination, optics, and video tube. Pixels corresponding to the cell perimeter (+ 5% of the cell radius) were sorted into 16 sectors according to their angle with respect to the electric field (Fig. 1, inset). The intensities in these sectors were used to estimate the distribution of fluorescently labeled sites around the cell perimeter. Control experiments using known dye concentrations have confirmed the validity of the data-gathering and -analysis techniques (Stollberg and Fraser, -1988). Theoretical considerations. Under “ideal” conditions (no interaction between the sites; spherical, nonconducting cells) the distribution of cell-surface sites at steady state in an electric field is a balance between electromigration and diffusion and has an analytical solution (Jaffe, 1977; Poo et al., 1979; McLaughlin and Poo, 1981; Ryan et al., 1988; Stollberg and Fraser, 1988). The form used here is that the site density as a function of B (C,) is C, =f,.a.exp([email protected] 0) + (1 -f,).C, where OL= fl.C,lsinh@), 0 = 1.5.E.r.mlD, 0 is the angular position relative to the anodal cell pole, r is the cell radius. C the initial molecular density, E the field strength, f, the fraction of ‘mobile sites, m the

electromigrational mobility constant, and D the diffusion constant. The 2 parameters m/D and fm were allowed to vary in fitting this description to data (all other parameters being measured quantities).

Results The experimental designutilizes enzymatic treatmentsand electric fields to study the migration and clustering of AChRs on cultured Xenopus myospheres.Following thesemanipulations, the distribution of fluorescently labeledmembranecomponents was quantified from digitized video images(see Fig. 1, inset, and Material and Methods). As a basisof comparison to enzymatically treated cells, untreated cultures were subjectedto electric fields and analyzed for the distribution of con A sites and AChRs. The results (Fig. 1) represent the 2 kinds of distributions documented previously (Stollberg and Fraser, 1988). The con A site distribution is as expected for “ideal” (noninteracting) sitesin that the data are well describedby a theoretical analysis which assumesthat electromigration and diffusion dominate the process(seeMaterials and Methods, Theoretical considerations).Furthermore, the sameparametersfit the steadystatecon A site distributions over a 4-fold rangein field strengths, and the distributions decay back to uniformity after termination of the field (Stollberg and Fraser, 1988). Thus, the con A sites

250

Stollberg

and

Fraser

* AChR

Accumulation

Does

Not Trigger

Clustering

A Cathode-facing 0 Anode-facing

Cl

L

60

30 Angle

90 from

Anode

120

150

180

(degrees)

Post-field

behave in accordance with the theoretical predictions under theseconditions. In contrast, the field-induced distribution of AChRs is markedly nonideal (Fig. 1). It deviates substantially from the bestminimum

at about

120” (Fig. 1, arrow). Moreover, the receptors continue to aggregateat the cathodal pole after termination of the field (Stollbergand Fraser, 1988). This finding, and the proximity of minimum receptor density to the forming aggregate,are consistent with the action of a specific diffusion trap for AChRs at the cathodal pole ofthe cell. AChRs clearly behavenonideally; their Table

1.

[Neurl”

N”

AChRs A IEElOr

0 0.05 0.2 1.0 2.0

(

0.43

60

(minutes)

is shown

in Figure 2A. Quantitative

analysis

of many suchimagesshowsclearly that the receptor distribution is bimodal, with increaseddensity at both the anodal and cathodal cell poles (Fig. 3). The frequency histogram of minimum

indices

-0.06 -0.13 0.00 -0.10

Reloxotion

60

distribution reflects a combination of electromigration, diffusion, and (predominantly) receptor aggregation(Stollberg and Fraser, 1988). The simplesthypothesis to explain receptor aggregation is that it is triggeredby the electromigrational increase in receptor density at the cathode-facing pole. To test the hypothesis that field-induced accumulation of AChRs causesthe receptor trap to form, cellswere subjectedto neuraminidasetreatments reported previously to reverse the electromigration of membrane components (Orida and Poo, 1978; McLaughlin and Poo, 1981). An example of a cell incubated with neuraminidase,placedin an electric field, and labeled with TMR-a-Bgt

Asymmetry

8 9 17 14

40

Figure 4. Density of AChRs at the 2 cell poles as a function of time, following termination of the electric field. Cells were incubated with 0.2 U/ml neuraminidase for 40 min, subjected to 8 V/cm for 40 min, and given the indicated times for postfield relaxation. Circles, Receptor density at the anode-facing cell pole. Dotted curve, Exponential with time constant 27.8 min, corresponding to D = 1.2 x 1O-9 cm*/sec (see Fig. 6, legend); X~,~, = 12. The rate of decrease in receptor density at the anode is consistent with the given diffusion constant. Triangles, Receptor density at the cathode-facing pole. Data are means f standard errors from 3-6 experiments, each consisting of 6-l 1 analyzed cells.

amongthecells.Dataaremeans? standarderrorsfrom 4 experiments, eachconsistingof 9-l I scanned cells.

curve, with a characteristic

20

0

Figure 3. Distribution of ACbRs following neuraminidase treatment and exposure to electric fields. Cells were treated with 0.2 U/ml neuraminidase for 40 min and subjected to a field of 8 V/cm for 80 min. The distribution is bimodal, indicating receptor accumulations at both anode- and cathode-facing poles. Solid curve, Theoretical fit for the sum of 2 ideally behaving populations: first population, m/D = 632 + 54 V-l, f, = 0.027 + 0.002; second population, m/D = -21 + 1 V-j, fm fixed at 0.97 (else the algorithm forced the value over 1.0); x2(,*, = 64. Dashed curve, Theoretical result fit by eye to left half of the data, yielding m/D = - 19 V-l (f, = 0.77, from data of Fig. 6). Dotted curve, Estimated distribution of AChRsresultingfrom aggregation alone-this curve is the difference between the solid and dashed curves (see Results). Inset, Frequency histogram showing the positions of minimal receptor density

fitting theoretical

pola pole

following

various

neuraminidase

A ‘eclanglc If: 0.01 + 0.05 + 0.06 5 0.00 f 0.02

0.24 -0.12 -0.20 0.01 -0.10

f f f f f

0.02 0.04 0.03 0.01 0.02

treatments

con A sites A IEClOr 0.46 f 0.02 0.13 + 0.04 0.10 f 0.02 0.13 * 0.01 0.07 + 0.03

A ‘cclanglc 0.41 zk 0.02 0.08 k 0.05 0.06 zk 0.02 0.09 k 0.02 0.06 f 0.03

Cells were incubated with the given concentrations of neuraminidase for 80 min and then subjected to fields of 8 V/cm for 40 min. The resulting distributions are summarized by the asymmetry index (A); A = (C, - C,)/(C, + C,), where C, and C, are the densities at the anodal and cathodal poles, respectively. A,,,,, is the asymmetry index calculated using is calculated using rectangles (cu. 3.5 x I8 pm) as the densities at sectors 1 and 16 as estimates for C, and C,. Am estimates for C, and C, and is presented to facilitate comparison with previously reported results. The 2 kinds of asymmetry indices give approximately the same values in the experiments summarized here. For both AChRs and con A sites, the indices are markedly reduced by neuraminidase treatment. The AChR distributions reverse in polarity by this measure, while con A site distributions approach neutrality. Values are means f standard errors and have been rounded to 2 decimal places. ” Neuraminidase concentration in units/ml. ” Number of cells analyzed. ‘ Same experiments as Figure 2; the cultures were labeled with 300 nM TMR-ol-Bgt (n = 7 cells) or 25 fig/ml TMR-con A (n = I I cells).

The Journal

60

30 Angle

90 from

Anode

120

150

0

180

30

60 Angle

(degrees)

Figure 5. Field-induced AChR distributions following progressively more extensive neuraminidase digestions. Cells were incubated with the indicated concentrations of neuraminidase for 80 min and then subjected to a field of 8 V/cm for 40 min. Although there may be some differences in the distributions, the significant features remain unaltered. All 3 distributions show the characteristic bimodality consistent with receptor electromigration toward the anode and receptor aggregation at the cathode. Data are means + standard errors for 6-9 cells per condition.

showsthat the bimodality of the AChR distribution is a feature common to most individual cells, rather than an artifact resulting from the averaging of 2 cell populations with density minima at opposite poles(Fig. 3, inset). The bimodality of this AChR distribution implies 2 opposed processes. The density profile near the anodal pole isreminiscent of an ideal distribution, while that near the cathodal pole suggeststhe operation of a diffusion trap (compare to Fig. 1). If this interpretation is correct, it shouldbe possibleto decompose the overall distribution into these 2 components. An estimate of the contribution made by ideal behavior is shown in Figure 3, dashedcurve. The difference between this estimate and the original distribution is shown by the dotted curve, which is similar to the AChR distribution (dominated by receptor aggregation) shown in Figure 1. This analysis strengthensthe interpretation that the data of Figure 3 represent the sum of 2 opposedprocesses:ideal electromigration (toward the anodal cell pole) and field-induced aggregationof receptors(at the cathodal pole). To further test this interpretation, the AChR distribution on neuraminidase-treated cells was monitored following a postfield relaxation period (Fig. 2B). Resultsaveragedfrom several such experiments show that the AChR density continued to increaseonly at the cathodal pole, as in cells not treated with neuraminidase(Fig. 4). Thus, the AChR aggregationevent is confined to the cathode-facing pole, consistent with the interpretation given the data of Figure 3. The averagereceptor density at the anode-facingpole wasaslarge asthat at the cathodefacing pole immediately following the field (Figs. 3,4). However, the anodal density decayed back to control levels after termination of the field (Fig. 4). The simplest interpretation of this result is that elevated AChR density in and of itself is not sufficient to induce receptor aggregation(seealso Discussion). Another possibleexplanation for the data of Figures 3 and 4 would be that the conditions of neuraminidase treatment resulted in a classof digestedreceptors that is incompetent with respectto aggregationand an undigestedclassthat is still able to aggregate.This could be attributed to the action of neur-

of Neuroscience,

January

120

90 from

Anode

1990,

150

70(l)

251

160

(degrees)

Figure 6. Steady-state distributions of AChRs following treatment with neuraminidase and trypsin. Cells were incubated with 0.2 U/ml neuraminidase for 40 min, 0.1% trypsin for 20 min, and then exposed to electric fields for 40 min. Circles, 4 V/cm. Triangles, 8 V/cm. Curves, Theoretical fit to ideal distributions; m/D = -34 k 4 V-‘,.f, = 0.77 f 0.08,x2,29, = 117. The same parameters fit the 2 distributions reasonably well, thouah the nossibilitv of nonideal behavior at the hiaher field strength cannot be eliminated (see Discussion). Data are means f standard errors from 3-4 experiments, each consisting of 5-12 analyzed cells.

density location

aminidase itself or to a contaminant activity in the enzyme preparation. According to this “partial digestion” scenario,the elevated receptor density at the anodal pole (Fig. 3) consistsof the digestedreceptors,which consequentlyfail to aggregateafter

I 20

0 l

Duration A Post-field

40

60

80

of Field (minutes) Relaxation (minutes)

Figure 7. Change in AChR asymmetry as a function of time. The asymmetry index (A) is determined by the densities at the anodal (C,) and cathodal (C,) cell poles and is defined as A = (C, - C,)/(C, + C,). Densities at sectors 1 and 16 were used to estimate C, and Cc. Cells were incubated with 0.2 U/ml neuraminidase for 40 mitt, 0.1% trypsin for 20 min, and then subjected to fields of 8 V/cm. Circles, Asymmetry development with time in the field. Cells were subjected to the indicated field duration and analyzed for sector asymmetry. Curve, Best-fit exponential to the data; time constant = 24 k 5 min, x2,,,= 2. The receptor diffusion constant (D) = r* (1 - 0.1.p2)/2.r (Poo, 1982). Given m/D = -34 V-l (Fig. 6, legend), r = 20 pm, and E = 8 V/cm, @is approximately -0.8. Together with the observed T, this gives an estimate of D = 1.3 (-c 0.2) x lo-‘+ cmZ/sec. Triangles, Asymmetrydecaywith time postfield. Cells were subjected to the field and then aiven the indicated times for postfield relaxation. Curve, Best-fit exponential to the data: time constant = 31 + 5 min, x2,,, = 6. The decay of asymmetry is approximated bv an exponential such that D = rY2.r (Huana. 1973). This leads to an estimate for D of = 1.1 (kO.2) x lo-; cm&&. Data are means + standard errors from 3-4 experiments, each consisting of 7-l 1 analyzed cells.

252

Stollberg

and

Fraser

- AChR

Accumulation

Does

Not Trigger

Clustering

field termination (Fig. 4). If the standard incubation conditions result in a partial digestion phenomenon as outlined above, then more extensive digestion should eliminate all aggregation activity, resulting in a distribution that is ideal and maximal at the anodal cell pole. As shown in Figure 5, such distributions were not seen. Instead, incubation conditions that should yield 20 times the digestion of the standard conditions result in bimodal distributions, showing that AChRs still aggregate at the cathodal cell pole. The marginal changes in the field-induced distribution of AChRs with the different neuraminidase concentrations (Fig. 5) suggest that the standard incubation conditions are sufficient to remove nearly all of the susceptible negative charges from the cell surface. In support of this, the electromigration of con A binding sites, which is altered by the standard incubation conditions, is not altered further by more extensive neuraminidase digestion (Table 1). These results argue strongly that partial digestion of receptors by the neuraminidase preparation is not responsible for the failure of AChRs to aggregate at the anodal pole. AChR aggregation, seen as a consequence of exposure to electric fields, can be blocked by mild digestion with trypsin (Stollberg and Fraser, 1988). If the data of Figure 3 represent the opposed actions of electromigration and receptor aggregation, blockage of receptor aggregation by trypsin treatment should result in an ideal receptor distribution with a maximum at the anodal cell pole. This prediction is borne out by experiments in which cells were treated with neuraminidase and trypsin before field application (Figs. 2C, 6). These distributions are well fit by the theoretical predictions for ideally behaving sites (Fig. 6, legend). The distributions support the conclusion that the direction of AChR electromigration has indeed been reversed by neuraminidase treatment. Because AChRs behave ideally after treatment with neuraminidase and trypsin, the development and decay of asymmetry can be used to estimate the receptor diffusion constant (D). Rates of development and decay of AChR asymmetries are shown in Figure 7. These time courses are well fit by exponentials indicating that D = 1.3 x 1O-9 cm2/sec (asymmetry development), and D = 1.1 x 1Om9cm2/sec (asymmetry decay). The agreement between these 2 estimates is noteworthy and is consistent with the characterization of AChR behavior as ideal following incubation with neuraminidase and trypsin. Discussion This study focuses on the initial AChR clustering events in a model system-cultured Xenopus myospheres exposed to electric fields. Receptor aggregation is suggested by the nonideal behavior of AChRs in response to such fields (Fig. 1). Other experiments have shown that AChRs continue to aggregate at the cathode-facing pole after the field is terminated (Stollberg and Fraser, 1988). These results are consistent with a slight electromigrational accumulation of receptors at the cathodefacing pole, superimposed on a much larger accumulation due to specific receptor trapping. The trap is selective for receptors and is triggered in the absence of neurons, surface contacts, or exogenous tissue-derived factors, rendering this a useful model system in which to study mechanisms of receptor aggregation on the molecular level. In the present work we have tested a simple and attractive hypothesis to explain the field-induced triggering of AChR aggregation (there are only 2 other possibilities-see below). The hypothesis holds that receptor aggregation is triggered by locally

increased receptor density, due in these experiments to electromigration (Poo, 1982; Stollberg and Fraser, 1988; for theoretical framework see Gershon, 1978). This view is consistent with the observation that transient, local increases in receptor density precede the sorting out of receptors into a “lattice” arrangement during cluster formation at myotube-substrate contacts (Pumplin and Bloch, 1987) and has been incorporated into a scheme accounting for the contact-mediated induction of receptor clusters (Bloch and Pumplin, 1988). The hypothesis is also consistent with the much-studied phenomenon of capping, which is caused by the externally induced rearrangement (patching) of cell surface molecules by multivalent ligands (Taylor et al., 197 1; de Petris and Raff, 1973; for review see Bourguignon and Bourguignon, 1984). Other experimental approaches to receptor clustering have suggested a requirement for complex interactions with other molecules, which might appear to rule out the hypothesis under test here. In particular, there is considerable evidence for the involvement of cytoskeletal anchoring of AChR clusters some hours after formation (Bloch and Hall, 1983; Bloch, 1986; Podleski and Salpeter, 1988). However, there is no compelling evidence for the causal role of these connections in initial clustering events (see Kuromi et al., 1985). Moreover, it has been shown previously that cytoskeleton-disrupting drugs do not reduce the field-induced aggregation of receptors in Xenopus myosphere cultures (Orida and Poo, 1978, 1980; Stollberg and Fraser, 1988). Finally, we know that cytoskeletal anchoring cannot be the initial clustering event at nerve-muscle contacts in vivo. This is because the signal for receptor clustering arises from interactions between the muscle cell and motor neuron and must therefore be transduced by some combination of binding by tissue-derived factors (e.g., Godfrey et al., 1984; Usdin and Fischbach, 1986), contact-mediated events (Peng et al., 1981; Bloch and Pumplin, 1988) or endogenous electric fields (Fraser and Poo, 1982). Neuraminidase, which removes sialic acid residues and alters the surface charge of the cell, has been reported to reverse the field-induced migration of AChRs and con A sites (Orida and Poo, 1978; McLaughlin and Poo, 198 1). This suggests a means to test the experimental hypothesis directly, by determining whether such conditions reverse receptor electromigration, receptor aggregation, or both. The experiments described here indicate that neuraminidase-treated cultures do show reversed electromigration of AChRs, but that receptor aggregation continues to take place at the cathodal pole, as in control (enzymatically untreated) cultures. The distribution of AChRs, following neuraminidase treatment and exposure to electric fields, is bimodal; the density is elevated at both cell poles (Fig. 3). This is a property of most individual cells, rather than a consequence of 2 or more cell populations (Fig. 3, inset). The observation of a bimodal distribution is a novel finding and depends on the spatially resolved quantitation of receptor density employed in these experiments. Earlier experiments, in which distributions were characterized by an asymmetry index based solely on density at the cathodal and anodal cell poles, emphasized the reversal of field-induced asymmetry following neuraminidase digestion (Orida and Poo, 1978; McLaughlin anad Poo, 1981). We find that the receptor asymmetry reverses, as shown in Table 1, in accord with the previous work. The con A site asymmetry decreased to near zero but did not actually reverse in response to neuraminidase digestion (Table 1). A difference as small as 1O-20% in the charge susceptible to neuraminidase cleavage could account for this

The Journal

minor discrepancy. Given the possible differences in cultures and reagents, this small difference between our results and those reported previously is probably not important. After termination of the field, neuraminidase-treated cultures display continued receptor aggregation at the cathodal cell pole, but the elevated receptor density at the anodal pole decreases (Fig. 4). This suggests that elevated receptor density is not sufficient to trigger receptor aggregation. It might be thought that partial digestion could account for this result if more digested receptors are not competent with respect to aggregation. The argument would hold that native AChRs electromigrate to the cathodal cell pole and are triggered to aggregate by their elevated density there; in contrast, digested receptors electromigrate to the anodal pole and cannot aggregate. It is difficult to reconcile this scenario with the charge-removing effects of neuraminidase; receptors that have been more extensively cleaved of sialic acid residues will be less negatively charged and will therefore migrate preferentially toward the cathodal cell pole compared to native AChRs. Moreover, the partial digestion scenario is inconsistent with the finding that 20 x greater digestion with neuraminidase still yields a bimodal field-induced distribution of AChRs (Fig. 5, Table 1). Thus, the results cannot be explained by invoking partial digestion of receptors, whether by the neuraminidase itself or by a contaminant in the preparation. Treatment of cells with trypsin has been shown to block receptor aggregation, and to have little if any effect on the electromigration of receptors (Stollberg and Fraser, 1988). If the bimodal receptor distributions in neuraminidase-treated cells reflect the opposition of electromigration and receptor aggregation, cells incubated with neuraminidase and trypsin should show ideal AChR distributions with the maximal density at the anodal cell pole. In agreement with this expectation, the distributions seen under these conditions are consistent with the predictions of ideal electromigration (Fig. 6). The distribution seen at 8 V/cm is slightly less asymmetric than would be predicted on the basis of the 4 V/cm data, which may reflect the presence of nonideal interactions between receptors at the higher field strength (Ryan et al., 1988; Stollberg and Fraser, 1988). These results support the view that neuraminidase treatment reverses electromigration of receptors, so that receptor density is electromigrationally lowered at the cathodal cell pole. Thus, the formation of receptor aggregates at the cathodal pole after neuraminidase digestion indicates that elevated receptor density is not necessary to trigger receptor aggregation. Because treatment of the cells with trypsin creates ideally behaving receptors, the rates at which asymmetries develop during field administration, and decay after field termination, can be used to estimate the receptor diffusion constant. In principle, such kinetic studies could be performed with trypsin treatment alone; however, the electromigration of AChRs under these conditions is too small to permit accurate measurements (Stollberg and Fraser, 1988). This limitation is circumvented by combined neuraminidase and trypsin treatments, which result in an easily measurable electromigration of AChRs. The results of such experiments provide an estimated diffusion constant of about 1.2 x 1O-9 cmZ/sec (Fig. 7, legend). Although the AChRs are 2 enzymatic steps removed from native receptors, the alteration of receptor diffusion is probably minimal. Trypsin digestion leaves the function and ol-Bgt binding of the receptor intact (Stollberg and Fraser, 1988) and digestion with other proteases reveals that AChRs are resistant to changes in function, size, shape, and sedimentation under nondenaturing con-

of Neuroscience,

January

1990,

IO(l)

253

ditions (Lindstrom et al., 1980). Neuraminidase digestion had no effect on cu-Bgt binding (data not shown) and should not materially affect the receptor mass. Therefore, the AChR diffusion constant measured in these experiments is probably close to that for native receptor. Estimates of the AChR diffusion rate are critical to an understanding of the molecular mechanisms responsible for receptor clustering. In particular, the diffusion trap hypothesis assumes that receptor diffusion is rapid enough to permit significant numbers of AChRs to be trapped within their average lifetime (Fraser and Poo, 1982; Poo, 1982; Kuromi et al., 1985; Stollberg and Fraser, 1988). The diffusion constant reported here (D = 1.2 x 10m9cm2/sec) is easily large enough to account for the receptor aggregation seen in Xenopus myosphere cultures (Stollberg and Fraser, 1988). The value is similar to that found for AChRs on cultured Xenopus muscle cells (2.6 x 10m9cm*/ set; Poo, 1982) and for extrajunctional AChRs in Xenopus tadpole muscle (1.5-4 x 1O-9 cm2/sec; Young and Poo, 1983) as measured by the diffusion of functional AChRs into a region of toxin-inactivated receptors. It is also similar to estimated diffusion constants for plasma membrane antigens on cultured embryonic muscle cells as measured by the diffusion of a locally labeled population (l-3 x 1Om9cm2/sec; Edidin and Fambrough, 1973), and to that of con A sites on cultured Xenopus muscle cells (2.3 x 1O-9 cm2/sec; Stollberg and Fraser, 1988). However, our estimate is considerably higher than the receptor diffusion constant determined by photobleaching recovery methods for Xenopus muscle cultures (2.5 x lo-lo cm2/sec; Kuromi et al., 1985) and cultured rat myotubes (5 x 10-l’ cm21 set; Axelrod et al., 1976). It has been suggested that diffusion estimates based on photobleaching may be smaller than those based on the decay of a gradient because intermolecular interactions lead to different rates of “self’ vs “mutual” diffusion (Scalettar et al., 1988). Our diffusion measurements, performed in the presence of gradients, would presumably reflect a measure of mutual diffusion. However, the estimates that are in agreement with our own (Edidin and Fambrough, 1973; Poo, 1982; Young and Poo, 1983) were obtained in the absence ofgradients and are therefore measures of self-diffusion. Thus, it appears that the distinction between mutual and self-diffusion cannot account for the discrepancy between our values and those obtained by photobleaching methods. The triggering of receptor aggregation The simple hypothesis that a homogeneous AChR population is triggered to aggregate by increased local density is excluded by the results discussed above. The data of Figure 4 show that AChR density at the anodal pole is elevated immediately after field termination, but decreases during the postfield period. Thus, elevated receptor density is not sufficient to induce aggregation. Furthermore, receptor aggregation is initiated at the cathodefacing pole despite the electromigrational lowering of AChR density (evident when receptor aggregation is blocked; Fig. 6). Therefore, local elevation of receptor density is not necessary to trigger receptor aggregation. There remain only 2 classes of mechanisms for the triggering of receptor aggregation that are consistent with our results. First, receptor aggregation may be triggered by the electromigrational accumulation of some molecule(s) as yet not probed for in this model system. Work underway in several laboratories is directed toward characterization of tissue-derived factors that induce AChR aggregates in vitro and may play a role in AChR clustering

254

Stollberg

and

Fraser

* AChR

Accumulation

Does

Not Trigger

Clustering

in vivo (e.g., Godfrey et al., 1984; Usdin and Fischbach, 1986; for review see Schuetze and Role, 1987). The in vitro action of such factors clearly requires interaction with cell-surface components; an attractive possibility is that the same components are responsible for the accumulation documented here. Another possibility is that the “other molecules” are in fact a distinct receptor subpopulation. This seems unlikely, as it would require significant differences in the properties of receptor subtypes (i.e., aggregation potential, extracellular charge, susceptibility to neuraminidase digestion), but the possibility cannot be rigorously excluded. A second class of mechanism for the triggering of receptor aggregation is suggested by the depolarization of the cell membrane that is associated with the application of electric fields. At 8 V/cm, a myosphere with a radius of 20 km will be depolarized by about 16 mV at the cathode-facing pole. Thus, voltage-sensitive mechanisms must be considered a possible trigger for receptor aggregation. The data could be explained, under such a scenario, if the adhesion or cohesion events responsible for receptor aggregation were controlled in a voltage-dependent manner. We attempted to distinguish between the 2 hypotheses (molecular accumulation vs voltage-dependent action) by searching for neuraminidase treatment conditions under which receptor aggregation is reversed (i.e., aggregation occurs at the anodal pole). As such treatments affect the electromigrational driving force, but not local membrane potentials, reversed receptor aggregation would rule out the direct involvement of voltagesensitive mechanisms. We have examined AChR distributions in cells incubated at up to 2 U/ml neuraminidase without observing such a reversal of aggregation (Table 1). It may be that the extracellular charge on the molecules that trigger receptor aggregation is such that even extensive treatment cannot cause reversal of their electromigration; hence this negative result does not distinguish between the 2 hypotheses. Accordingly, experiments based on other distinct predictions made by the 2 hypotheses must be performed. The current experimental system is ideally suited to further examination of the molecular mechanisms of receptor aggregation by virtue of the simplicity and rapid response of the system, as well as the quantitative nature of the analysis. Summary of conclusions Receptor aggregation can be initiated independently of cell-substrate contacts and soluble factors in a simple model system. This indicates that the system preserves some endogenous, surface-associated components that are sufficient to induce local, specific, receptor clustering. Neuraminidase treatment reverses the electromigration of AChRs, which is consequently directed toward the anodal pole, but leaves unchanged the receptor aggregation event at the cathode-facing pole. After termination of the electric field, receptor density in these cultures decreases at the anodal pole, but continues to increase at the cathodal pole. Therefore, the local accumulation of AChRs is neither necessary nor sufficient to trigger the receptor aggregation event. With the elimination of this hypothesis, only 2 possible causes of receptor aggregation in this model system remain: (1) field-induced accumulation of some other molecule(s), or (2) a voltage-sensitive mechanism.

References Anderson, M. J., and M. W. Cohen (1977) Nerve-induced and spontaneous redistribution of acetylcholine receptors on cultured muscle cells. J. Physiol. Lond. 268: 757-73. Axelrod, D., P. Ravdin, D. E. Koppel, J. Schlessinger, W. W. Webb, E. L. Elson, and T. R. Podleski (1976) Lateral motion of fluorescently labeled acetylcholine receptors in membranes of developing muscle fibers. Proc. Natl. Acad. Sci. USA 73: 4594-4598. Blackshaw, S., and A. Warner (1976) Onset ofacetylcholine sensitivity and endplate activity in developing myotome muscles of Xenopus. Nature 262: 2 17-2 18. Bloch, R. J. (1986) Actin at receptor-rich domains of isolated acetylcholine receptor clusters. J. Cell Biol. 102: 1447-1458. Bloch, R. J., and Z. W. Hall (1983) Cytoskeletal components of the vertebrate neuromuscular junction: Vinculin, o-actinin, and filamin. J. Cell Biol. 97: 217-223. Bloch, R. J., and D. W. Pumplin (1988) Molecular events in synaptogenesis: Nerve-muscle adhesion and postsynaptic differentiation. Am. J. Physiol. 254: c345-c364. Bourguignon, L. Y. W., and G. J. Bourguignon (1984) Capping and the cvtoskeleton. Int. Rev. Cytol. 87: 195-224. Creazzo, T. L., and G. S. Sohai (1983) Neural control of embryonic acetylcholine receptor and skeletal muscle. Cell Tissue Res. 228: l12. Dennis, M. J. (1981) Development of the neuromuscular junction: Inductive interactions between cells. Annu. Rev. Neurosci. 4: 43-68. de Petris, S., and M. C. Raff (1973) Normal distribution, patching and capping of lymphocyte surface immunoglobulin studied by electron microscopy. Nature (Lond.) New Biol. 241: 257-259. Edidin, M., and D. Fambrough (1973) Fluidity of the surface of cultured muscle fibers. J. Cell Biol. 57: 27-37. Edwards, C., and H. L. Frisch (1976) A model for the localization of acetylcholine receptors at the muscle endplate. J. Neurobiol. 7: 377381. Frank, E., and G. D. Fischbach (1979) Early events in neuromuscular junction formation in vitro. Induction of acetylcholine receptor clusters in the postsynaptic membrane and morphology of newly formed nerve-muscle synapses. J. Cell Biol. 83: 143-58. Fraser, S. E., and M.-M. Poo (1982) Development, maintenance, and modulation of patterned membrane topography: Models based on the acetylcholine receptor. Cur. Top. Dev. Biol. 17: 77-100. Gershon, N. D. (1978) Model for capping of membrane receptors based on boundarv surface effects. Proc. Natl. Acad. Sci. USA 75: 1357-1360. Godfrey, E. W., R. M. Nitkin, B. G. Wallace, L. L. Rubin, and U. J. McMahan (1984) Comnonents of Toruedo electric organ and muscle that cause aggregation of acetylcholine- receptors on cultured muscle cells. J. Cell Biol. 99: 6 15-627. Huang, N. W. (1973) Mobility and diffusion in the plane of cell membrane. J. Theor. Biol. 40: 11-16. Jaffe, L. F. (1977) Electrophoresis along cell membranes. Nature 265: 600-602. Kidokoro, Y., M. J. Anderson, and R. Gruener (1980) Changes in synaptic potential properties during acetylcholine receptor accumulation and neuronspecific interactions in Xenopus nerve-muscle cultures. Dev. Biol. 78: 464-483. Kuromi, H., and Y. Kidokoro (1984) Nerve disperses preexisting acetylcholine receptor clusters prior to induction of receptor accumulation in Xenopus muscle cultures. Dev. Biol. 103: 53-6 1. Kuromi, H., B. Brass, and Y. Kidokoro (1985) Formation of acetylcholine receptor clusters at neuromuscular junction in Xenopus cultures. Dev. Biol. 109: 165-176. Lindstrom, J., W. Gullick, B. Conti-Tronconi, and M. Ellisman (1980) Proteolytic nicking of the acetylcholine receptor. Biochemistry 19: 479 14795. Luther, P. W., and H. B. Peng (1985) Membrane-related specializations associated with acetylcholine receptor aggregates induced by electric fields. J. Cell Biol. 100: 235-244. McLaughlin, S., and M.-M. Poo (1981) The role of electro-osmosis in the electric-field-induced movement of charge macromolecules on the surfaces of cells. Biophys. J. 34: 85-93.

The Journal

Nieuwkoop, P. D., and J. Faber (1962) Normal Table of Xenopus laevis, North Holland, Amsterdam. Orida N., and M.-M. Poo (1978) Electrophoretic movement and localization of acetylcholine receptors in the embryonic muscle cell membrane. Nature 275: 31-35. Orida, N., and M.-M. Poo (1980) On the developmental regulation of acetylcholine receptor mobility in the Xenopus embryonic muscle membrane. Exp. Cell Res. 130: 28 l-290. Orida, N., and M.-M. Poo (198 1) Maintenance and dissolution of acetylcholine receptor clusters in the embryonic muscle cell membrane. Dev. Brain Res. I: 293-298. Peng, H. B., P.-C. Cheng, and P. W. Luther (198 1) Formation of ACh receptor clusters induced by positively charged latex beads. Nature (Lond.) 292: 83 l-834. Podleski, T. R., and M. M. Salpeter (1988) Acetylcholine receptor clustering and triton solubility: Neural effect. J. Neurobiol. 29: 167185. Poo, M.-M. (1982) Rapid lateral diffusion of functional ACh receptors in embryonic muscle cell membrane. Nature 295: 332-334. Poo, M.-M., and K. R. Robinson (1977) Electrophoresis of concanavalin A receptors along embryonic muscle cell membrane. Nature 265: 602-605. Poo, M.-M., W-J. H. Poo, and J. W. Lam (1978) Lateral electrophoresis and diffusion of concanavalin A receptors in the membrane of embryonic muscle cell. J. Cell Biol. 76: 483-50 1. Poo, M.-M., J. W. Lam, N. Orida, and A. W. Chao (1979) Electrophoresis and diffusion in the plane of the cell membrane. Biophys. J. 26: l-22. Pumplin, D. W., and R. J. Bloch (1987) Disruption and reformation of the acetylcholine receptor clusters of cultured rat myotubes occur in two distinct stages. J. Cell Biol. 104: 97-108.

of Neuroscience,

January

1990,

IO(l)

255

Role, L. W., V. R. Matossian, R. J. O’Brien, and G. D. Fischbach (1985) On the mechanism of acetylcholine receptor accumulation at newly formed synapses on chick myotubes. J. Neurosci. 5: 2197-2204. Role, L. W., D. Roufa, and G. D. Fischbach (1987) The distribution of acetylcholine receptor clusters and sites of transmitter release along chick ciliary ganglion neurite-myotube contacts in culture. J. Cell Biol. 104: 371-379. Ryan, T. A., J. Myers, D. Holowka, B. Baird, and W. W. Webb (1988) Molecular crowding on the cell surface. Science 239: 6 l-64. Scalettar, B. A., J. R. Abney, and J. C. Owicki (1988) Theoretical comparison of the self diffusion and mutual diffusion of interacting membrane proteins. Proc. Natl. Acad. Sci. USA 85: 6726-6730. Schuetze, S. M., and L. W. Role (1987) Developmental regulation of nicotinic acetylcholine receptors. Annu. Rev. Neurosci. 10: 403-457. Stollberg, J., and S. E. Fraser (1988) Acetylcholine receptors and concanavalin A-binding sites on cultured Xenopus muscle cells: Electrophoresis, diffusion, and aggregation. J. Cell Biol. 107: 1397-1408. Taylor, R. B., P. H. Duffus, M. C. Raff, and S. de Petris (1971) Redistribution and pinocytosis of lymphocyte surface immunoglobulin molecules induced by anti-immunoglobulin antibody. Nature (Lond.) New Biol. 237: 225-229. Usdin, T. B., and G. D. Fischbach (1986) Purification and characterization of a polypeptide from chick brain that promotes the accumulation of acetylcholine receptors in chick myotubes. J. Cell Biol. 103: 493-507. Young, S. H., and M.-M. Poo (1983) Rapid lateral diffusion of extrajunctional acetylcholine receptors in the developing muscle membrane of Xenopus tadpole. J. Neurosci. 3: 225-23 1. Ziskind-Conhaim, L., I. Geffen, and Z. W. Hall (1984) Redistribution of acetylcholine receptors on developing rat myotubes. J. Neurosci. 4: 2346-2349.